Views
3 years ago

MINING SOCIAL NETWORKS AND THEIR VISUAL SEMANTICS ...

MINING SOCIAL NETWORKS AND THEIR VISUAL SEMANTICS ...

10 Michel Plantié,

10 Michel Plantié, Michel Crampesnormalized to 1 on the y-coordinates. When looking at this figure, the proximity forceand the simple force behave very much the same as far as the number of edges isconcerned. Few couples are captured before the value 900 of the distance. The cohesionforce is much flatter and it captures more couples than the two other forces at smallerdistances. The choice of a significant number of edges is also partly given by the referentgraph in Figure 1 because this is how the civil semantics was manually defined with 41edges. Taking into account these two points of view we chose 33 as the right number ofedges for the three different networks and took the corresponding distance thresholdvalue for each graph. Figures 4, 5 and 6 give the final displays for the three forcesreduced to 33 edges.hal-00659783, version 1 - 13 Jan 2012Fig. 3. Number of edges vs. distance for the 3 socialforces.4.4. Semantic analysis of the networksFig. 4. Reduced self-displayed simple force graph(33 edges)Figures 4, 5 and 6 show two remarkable properties: i) the simple force and the proximitygraphs look very much the same and ii) they have a different but very similar topology(Figure 4 and 6). Roughly 40% of the people are disconnected and the remaining people,mostly the close family, gather around the bride and the groom like a starfish.Semantically, these two social forces extract networks that bring to light the deepmeaning of the marriage: the union of two families through the union of two people.Fig. 5. Reduced self-displayed cohesion graph(33 edges)Fig. 6. Reduced self-displayed proximity graph(33 edges)

hal-00659783, version 1 - 13 Jan 2012Mining social networks and their visual semantics From social photos 11More precisely, the simple force considers the number of times a couple appears ondifferent photos. This social force takes into account the number of object-concepts(groups of photos with the same people). Obviously the bride and the groom are theheroes: their nodes are very close and most of the liaisons start from them. Members ofthe close family are most of the time captured with them (links only with the bride andthe groom) and sometimes together either with or without the bride and the groom. Theother people are passive participants that are less often on photos and consequently showno liaisons (at least on the reduced graph which is not the case on the original graph).With the proximity force the more a couple is isolated on a photo, the stronger it is. Acouple lost among many other people is less prone to social relations. It is surprising thatthe resulting network looks very much the same as the simple force network. This is aresult which was not expected. It suggests the idea that a strong couple is bound to bepresent on a photo on its own or with few people. It can be noticed that both forces takeinto account the number of concepts as their denominator.The cohesion network is very different. This force does not take into account the numberof object-concepts. A couple is strong if the two persons are present together on thephotos and rarely separated. In the resulting reduced graph in Figure 5 only three peopleare disconnected. The graph is flat with long chains of nodes. The heroes of the wedding,the bride and the groom, are still there with the most important node degrees. They nowclearly belong to two small cliques which are their respective close families. The otherparticipants have other relations with other people and as a result become side actors ofthe ceremony. The big difference with the referent network is that in this social networkthe human relations are more important than the civil relations. It captures individualencounters and the participants may remember whom she or he talked to or was seatedby. There are only three disconnected people and they look like they were lost. Actuallythey were the less implicated people in the wedding. It would be good to go back andreintroduce some edges from the original network to reintegrate them if the network hadto be shown to the participants. Semantically the cohesion graph is a partial mirror of animportant moment of social life for all the participants.5. Photo diffusion policy for the wedding eventIn Section 3 a photo dissemination policy was presented. In this section this policy isapplied to the ground truth example of the wedding. The method consists of analyzing theoriginal and reduced graphs and extracting tribes. Photos that are the closest to each tribeafter application of the Jaccard distance are prone to be sent to all the members of thecorresponding tribe.5.1. Tribe mining, two strategiesTo compute tribes we used the reduced graphs previously described. Each force producesa different graph. As was already exposed, a reduced graph for a particular force iscomputed by keeping edges whose length (the distance between the two nodes connectedby the corresponding edge) is under a chosen threshold. The threshold is chosen to get agraph with 33 edges.5.1.1. Tribe mining - First method: extract all tribes and decideWe can compute all sub-graphs from the 33 edges reduced graphs. The computation ofall sub graphs is still time consuming with the possibility of generating 2 33 sub-graphs in

Phylogenetic Networks - Visualization Techniques for ... - DIMACS
Semantic Adaptivity and Social Networking in Personal Learning ...
Data Mining Versus Semantic Web
Data mining and visualization
Visual Data Mining and Document Collections Visualization - USP
The Visual Worlds of Social Network Sites - Nomos
Data Mining Data Mining Presentation
Large Network Analysis and Visualization - Cyberinfrastructure for ...
Introduction to Text Mining and Semantics - Alta Plana
DIY Data Mining, Information Visualization, and Science Maps
Web Mining and Social Networking: Techniques and ... - tud.ttu.ee
Mining, Mapping, and Accelerating Scholarly Networks - Maryland ...
Data Visualization and Data Mining - Utah State University
A Pixel-Based, Semantically Rich Visualization Of Methods - ESUG
Semantic Sensor Networks: The W3C SSN-XG Ontology and How to ...
(3) Applications of Visual Data Mining - Utah State University
Statistical Graphics & Visual Data Mining in the Medical Field
Data Visualization in Data Mining - Chemical Engineering and ...
Jan Kleinnijenhuis (VU) Semantic network analysis: old ideas and ...
Security & Privacy in Online Social Networks - The Computer ...
analyzing and visualizing correspondence networks for browsable ...
MR SAS 8708ELP Product Brief - Visual Business NETwork
Mobile phone use and social network development ... - LIRNEasia
Visual Data Mining of Remote Sensing Data - Utah State University
Social Network Analysis - the Center on Early Adolescence